TOMLDecoder Is Now Faster Than C (Thanks to AI)

1 comments

• hairband_dude 2 hours ago
  Interesting. I'm curious if it did any SIMD tokenisation optimisations a la simdjson (not sure if swift has a SIMD API for it to use).

  It's also not obvious what optimisations the c/c++ libraries were compiled with; Should be -O3 -march=native at the very least.